Classroom Indoor Air Quality Time Series — Bangladesh, 2024 – 2025
Description
This dataset contains indoor classroom air-quality measurements collected by a custom sensor node in Bangladesh from 2024-09-25 19:25 to 2025-04-07 20:25. The system was deployed in a university classroom measuring 12.30 m × 6.95 m × 3.10 m with a partial false ceiling at 2.90 m. The room has no operable windows or mechanical ventilation; outside air enters primarily through a 21.4 cm × 46.8 cm louvered grille in the door. A curtain-wall window and a wall-mounted AC unit are present; the AC was used intermittently. This minimally ventilated setting enables clear observation of CO₂ accumulation and particulate dynamics during occupancy. The dataset contains 52,259 records, representing 93.51% of the theoretical 5-minute schedule over the collection window. Several multi-hour gaps exist (e.g., 2025-03-22 03:30 → 2025-03-23 22:40, ~43.2 h), consistent with temporary power or network interruptions, not a change in the sampling methodology. Variables (units): Timestamp (local, Asia/Dhaka, UTC+06:00); CO2 (ppm); pm1, pm2_5, pm10 (µg/m³); air_temperature (°C); humidity (%); air_pressure (Pa); deviceId; payload_id. A machine-readable data dictionary "MendeleyData_data_dictionary.csv" is provided. Instrumentation & calibration: The node integrates an MH-Z19B NDIR CO₂ sensor, a PMS5003 optical particle counter for PM, an SHT35 for temperature and humidity, and a BMP280 pressure sensor. Each sensor undergoes a standard stabilization period after power-up, is checked for offset/span consistency against manufacturer specifications, and is periodically co-located with a reference monitor (IQAir AirVisual Pro) to verify PM₂.₅, CO₂, and temperature agreement. Data quality is maintained through routine inspection of time-series behavior, health/status flags, and scheduled co-location checks to detect drift over the deployment. Processing & quality control: Minimal yet transparent handling preserved raw integrity while ensuring usability. All data entries were chronologically sorted by timestamp and scanned for duplicate entries based on timestamp and device ID, with no duplicates detected. No resampling, interpolation, smoothing, or artificial gap-filling was applied at any stage. Instead, the dataset retains its original granularity, allowing downstream users to apply their own thresholds and filtering logic. Basic anomaly detection was conducted through descriptive checks without excluding such records from the dataset. Intended use & limitations: Suitable for analyses of indoor ventilation, occupancy proxies (CO₂), episodic PM behavior, diurnal patterns, and time-series modeling. However, as data is from a single room, it may not represent conditions in other indoor spaces. PM₂.₅ readings may be affected by humidity. The room's location is approximate to protect privacy. Further documentation: Classroom characteristics (Section 2.2) and sensor calibration & QA (Section 3.4) are detailed in the IUB repository: https://ar.iub.edu.bd/handle/11348/992
Files
Steps to reproduce
Hardware & sensors: Custom node with MH-Z19B (CO₂, NDIR), PMS5003 (PM), SHT35 (temperature/humidity), and a BMP 280 barometric pressure sensor; ATmega2560 and ESP8266 microcontroller. Deployment context: Closed university classroom. No mechanical ventilation; outside air via 21.4 cm × 46.8 cm louvered door grille. Curtain-wall window present. Wall-mounted recirculating AC used intermittently. Calibration & QA (before deployment): CO₂ (MH-Z19B): manual “fresh-air” baseline alignment (~400 ppm). PM (PMS5003): ≥30 s warm-up; HEPA-filtered zero check. T/RH (SHT35) & BMP280 : factory-calibrated; subset cross-checked (≈±0.2 °C, ±2 % RH); no further adjustment. Validation: co-located comparisons with IQAir AirVisual Pro. Logging configuration: Fixed 5-minute sampling interval. Store local timestamps (YYYY-MM-DD HH:MM:SS, no embedded offset), CO₂ (ppm), PM₁/PM₂․₅/PM₁₀ (µg/m³), temperature (°C), RH (%), pressure (Pa), deviceId, payload_id. Data collection window: 2024-09-25 19:25 → 2025-04-07 20:25 (Asia/Dhaka). Multi-hour gaps may occur due to power/network interruptions. Export & processing: Export raw logs to CSV. Sort by Timestamp. Check duplicates on (Timestamp, deviceId) → none retained. No resampling, smoothing, or gap-filling. Descriptive screens only (values kept): physical-bounds flags and “stuck-value” runs (≥30 consecutive identical samples). Save as device2_air_quality_2024-09-25_to_2025-04-07.csv. Time standard. Local Asia/Dhaka (UTC+06:00); Bangladesh has no DST. Users may convert to ISO-8601 with +06:00 or to UTC.
Institutions
- Independent University
Categories
Funders
- Fablab IUB; Independent University, Bangladesh (IUB)